THE d O U H N A L OF BIOLOGICAL CHEMISTRY Vol. 266, No. 26, Issue of September 15, pp. 17584-17591, 1991 Printed in U.S.A.

Tandem Mass Spectrometry Identifies Sites of Three Post-translational Modifications of Spinach Light-harvesting Chlorophyll Protein I1 PROTEOLYTIC CLEAVAGE, ACETYLATION, AND PHOSPHORYLATION*

(Received for publication, April 25, 1991)

Hanspeter MichelS, Patrick R. Griffin§, Jeffrey Shabanowitzg, Donald F. Hunts, and John Bennettll From the Biology Department, Brookhauen National Laboratory, Upton, New York 11973 and the §Chemistry Department, University of Virginia, Charlottesuille, Virginia 22901

The photosynthetic membranes of spinach (Spinacia oleracea L.) chloroplasts were incubated with [ Y - ~ ~ P ] ATP. When the thylakoid membrane kinase was acti- vated with light, the 25- and 27-kDa forms of the light- harvesting chlorophyll a/b protein (LHC 11) were phos- phorylated on their amino termini. Treatment of the membranes with proteinase K or thermolysin released phosphopeptides which were purified by ferric ion af- finity chromatography and reverse phase high per- formance liquid chromatography. Sequencing of the phosphopeptides was performed with tandem quadru- pole mass spectrometry. Three different phosphopep- tides Ac-RKTAGKPKT, Ac-RKTAGKPKN, and Ac- RKSAGKPKN originating from class I LHC I1 were examined after release by thermolysin. One phospho- peptide, Ac-RRTVKSAPQ, originating from class I1 LHC I1 was examined after release by proteinase K. Each of the four LHC I1 phosphopeptides was derived from the amino terminus of a distinct protein. Peptides were acetylated at their amino-terminal arginine and were phosphorylated on either threonine or serine in the third position. We conclude that proteolytic proc- essing of pre-LHC I1 occurs at a conserved methionyl- arginyl bond and is followed by amino-terminal acet- ylation of the arginine and nearby phosphorylation of the mature LHC 11. Eight different peptides were syn- thesized in acetylated and nonacetylated forms as sub- strates for the thylakoid membrane kinase. From a comparison of the kinetics of phosphate incorporation into the peptides, we conclude that basic residues on both sides of the phosphorylation site are important for enzyme recognition. Acetylation of the amino ter- minus is not required for phosphorylation.

* Work was carried out at the Brookhaven National Laboratory under the auspices of the United States Department of Energy, Division of Biological Energy. Research was supported by Grant 84- CRCR-1-1487 from the United States Department of Agriculture. Work at the University of Virginia was supported by Grant CHE- 8618780 from the National Science Foundation. Instrument devel- opment funds (to D. F. H.) from The Monsanto Co. are also gratefully acknowledged. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$Present address and to whom correspondence and reprint re- quests should be addressed: Chemistry Dept., University of Virginia, Charlottesville, VA 22901

ll Present address: International Centre for Genetic Engineering and Biotechnology, NII Campus, Shaheed Jeet Singh Marg, New Delhi-110067, India.

The most abundant proteins in the photosynthetic mem- branes of green plants are the LHC 11’ (1). They are encoded by a family of nuclear cab genes ( 2 ) and synthesized as cytoplasmic precursors prior to import into the chloroplast (3). LHC I1 cab genes have been divided into two subclasses. Class 1 cab genes contain no intron and code for mature proteins of about 27 kDa which contain the motif Arg-Lys- (Thr/Ser)-(2 residues)-Lys- near the amino terminus. Class I1 cab genes contain an intron and code for mature proteins of about 25 kDa which contain the motif Arg-Arg-(Thr/Ser)- (1 residue)-Lys- (2). After import into the chloroplast, the pre-LHC 11s undergo a series of changes including proteolytic processing, insertion into thylakoid membranes, association with pigments, and reversible phosphorylation (l) , but it is not clear whether processing must precede the other changes (4).

It is assumed that these postsynthetic modifications depend on specific sequences intrinsic to pre-LHC I1 molecules. We are interested in the sequences which specify postsynthetic processing and phosphorylation of LHC 11s. This interest arises from the importance of proteolytic processing in LHC I1 assembly and the importance of phosphorylation in regu- lating the distribution of light energy between PS I and PS I1 during photosynthesis ( 5 ) . Processing and phosphorylation occur within a few residues of each other and may therefore be specified by overlapping sequences. Pre-LHC 11s are proc- essed to their mature size through removal of amino-terminal extensions known as transit peptides (1,3). A protease capable of cleaving these molecules has been partially purified (6). The same protease seems able also to cleave the precursor of another major nuclear-encoded chloroplast protein, the small subunit of RBUp,Case, and may correspond to a processing protease reported earlier (7). However, the presumed cleavage sites of pre-LHC 11s have little sequence similarity with the cleavage sites of pre-small subunits (3, 8). This may reflect the fact that the amino termini of mature LHC 11s are also involved in reversible phosphorylation, and residues around the amino termini may need to be compatible with both processing and phosphorylation.

A phosphopeptide derived from one of the major LHC 11s of pea was identified by a combination of protein sequencing (9) and cDNA cloning (10). However, the exact sites of phosphorylation and processing, as well as the identity of the amino terminus, were not established. Sequencing of LHC I1

I The abbreviations used are: LHC 11, light-harvesting chlorophyll a/b protein 11; diuron, 3-(3’,4’-d~chlorophenyl)-l,l-dimethylurea; HPLC, high performance liquid chromatography; PMSF, phenyl- methylsulfonyl fluoride; PS, photosystem.

17584

Post-translational Modifications of LHC 11 17585

by Edman degradation is hampered by amino-terminal block- ing (9, 11, 12).

We have sequenced the phosphorylation sites of four PS I1 proteins of spinach (13, 14). Like LHC 11, they reside on the photosynthetic membrane, are substrates of a redox-con- trolled protein kinase, and are phosphorylated a t or very close t o their amino termini. A tryptic peptide was isolated from each protein by ferric ion affinity chromatography and reverse phase HPLC, but only one of the four could be sequenced by Edman degradation. It was derived from the 8-kDa product of the chloroplast psbH gene and contained HN2-alanyl-(0- phospho)threonyl as its first 2 residues (13). The other three phosphopeptides (from the 32-, 43-, and 34-kDa products of the psbA, -C and -D genes) proved to be blocked at their amino termini and were sequenced by tandem mass spectrom- etry. In each case, the amino-terminal residue proved to be N-acetyl-(0-phospho)threonine (14). To clarify the chemical nature of the amino termini of LHC 11, we have applied the same techniques to peptides derived from the LHC I1 of spinach. Four closely related but distinct gene products were detected, each one acetylated at the amino terminus and phosphorylated on the 3rd residue. The data enable us to draw conclusions about the sites of proteolytic processing, acetylation, and phosphorylation in chloroplasts.

To probe the specificity of the thylakoid-bound protein kinase, we have examined the ability of synthetic analogs of the above peptides to act as kinase substrates. The feasibility of using synthetic peptide analogs as LHC I1 substrates has already been established (15-17). We used the peptide analogs to assess the impact on phosphorylation of (i) the basic amino acids on the amino-terminal and carboxyl-terminal sides of the phosphorylation site, (ii) the position of the hydroxyamino acid residue, and (iii) amino-terminal acetylation.

EXPERIMENTAL PROCEDURES

Preparation of Membranes-Spinach leaves (50 g, 8-20 weeks old) were homogenized as described (18). The homogenate was filtered through four layers of cheesecloth followed by one layer of Miracloth (Calbiochem). Thylakoid membranes were sedimented (8000 X g, 12 min) and washed twice with 35 ml of 10 mM Tricine (pH 7.8),10 mM NaC1,lO mM MgCl2,O.l mM PMSF (15,000 X g, 5 min). For labeling, thylakoids were suspended in 0.1 M sucrose,lO mM Tricine (pH 7.8),10 mM MgCI,,O.l mM PMSF to 1 mg of chlorophyll/ml. Phosphorylation was done with 0.2 mM ATP in light (200 pE.rn-'.s"). After 15 min, NaF was added to a concentration of 20 mM, and phosphorylation was done for another 20 min. Membranes were also labeled with [y- ,'"P]ATP (200-300 pCi/Fmol). They were sedimented (20,000 X g, 5 min) and then washed once with 10 mM Tricine (pH 7.8),10 mM EDTA,2O mM NaF,O.l mM PMSF (20 min, 0 "C); once with 10 mM Tricine (pH 7.8),20 mM NaF,1 M NaBr (30 min, 0 "C); and twice with 0.5% NH,HCO:JO mM NaF. The final pellet was suspended in 0.5% NH4HCO:I,20 mM NaF prior to proteolytic digestion.

Proteolytic Treatment of Membranes-Thylakoid membranes (60 mg of chlorophyll) were incubated in 0.5% NH4HC01, 20 mM NaF at 1 mg of chlorophyll/ml a t room temperature in the dark. Digestion with thermolysin (0.2 mg/ml) was done for 3 h at room temperature, and the reaction was stopped by adding EDTA and NaCl to final concentrations of 5 mM and 100 mM, respectively. Digestion with proteinase K (0.1 mg/ml) was done for 48 h at room temperature. The reaction was stopped by adding PMSF and NaCl to final concen- trations of 0.1 mM and 100 mM, respectively. After digestion, the thylakoid membranes were removed (20,000 X g, 30 min), and the supernatant was lyophilized. Four ml of H 2 0 was added and brought to pH 3-4 with acetic acid. The acid-insoluble material was removed (20,000 X g, 5 rnin), and the supernatant containing peptides was desalted on Sephadex G-15 (1.2 X 30 cm) in 0.1 M acetic acid.

Isolation of Phosphopeptides-The peptide-containing fractions were applied to an iminodiacetyl-Sepharose column (1 X 10 cm), which had been saturated with FeR+ as described (13,14). The column was washed with 15 ml each of (i) 0.1 M acetic acid, (ii) 0.1 M acetic acid, 0.5 M NaCl (pH 5), and (iii) 2% NH4Ac. Phosphopeptides were eluted with 2% NH,Ac brought to pH 8.5 with NHZI, and the phos-

phopeptide-containing fractions were lyophilized. Reverse Phase HPLC-Separation of phosphopeptides was per-

formed with a Beckman binary HPLC system and a reverse phase column (C,,, Ultrasphere, 4.6 X 250 mm, 5-pm pore size). For sepa- ration of peptides generated by thermolysin, we used a linear gradient of (i) aqueous buffer containing 100 mM sodium phosphate (pH 3.1) with 0.1% (w/v) hexanesulfonic acid (sodium salt) and (ii) 60% (v/v) acetonitrile containing 40 mM sodium phosphate (pH 3.1) with 0.1% hexanesulfonic acid (sodium salt). Prior to injection, samples were dissolved in 250 p1 of solvent which contained 2% (w/v) hexanesul- fonic acid (sodium salt). Phosphopeptide-containing fractions were then lyophilized and desalted on the reverse phase column by using a linear gradient of (i) 0.1% trifluoroacetic acid (w/v) in H20 and (ii) 0.1% trifluoroacetic acid (w/v) in acetonitrile. For separation of peptides from the proteinase K digest, we used a linear gradient of (i) 0.1% (v/v) heptafluorohutyric acid in H,O and (ii) 60% (v/v) acetonitrile containing 0.04% (v/v) heptafluorobutyric acid. Prior to injection, samples were dissolved in 250 p l of H,O containing 2% heptafluorobutyric acid. No desalting was necessary when heptafluo- robutyric acid was used.

Amino Acid Analysis-After purification by affinity chromatogra- phy and HPLC, phosphopeptides were analyzed for amino acid com- position (19). Hydrolysis was carried out in 6 M HC1 a t 105 "C for 22 h. The values for threonine and serine were corrected for 5 and 10% loss, respectively.

Mass Spectrometry-Mass spectra were recorded on either a triple quadrupole instrument (Finnigan-MAT, San Jose, CA) (20) or on a quadrupole Fourier transform instrument (21). Use of these instru- ments for the identification of phosphorylation sites has been de- scribed previously (14, 22). Electrospray mass spectra were recorded on the TSQ-70 instrument equipped with the newly developed Fin- nigan electrospray source. The electrospray needle was operated with a voltage differential of 3-5 kV and a sheath flow of 5 pl/min of a 3:l mixture of methanol,0.5% acetic acid. Collision-activateddissociation experiments were conducted a t energies of 20-25 electron volts for doubly charged ions. Argon a t a pressure of 3.5 mtorr was employed as the collision gas. Microcapillary HPLC experiments were con- ducted with fused silica columns having an inside diameter of 75 p and a length of 75 cm. The last 10 cm of the column was filled with reverse phase C," packing material. Peptides were eluted with a gradient of 0-80% acetonitrile (0.5% acetic acid) over a 20-min period a t a flow of 1-2 pl/min.

Peptide Synthesis and Purification-Synthetic peptides bound to the solid resin were obtained from Multiple Peptide Systems. Peptides were cleaved from the resin by treatment with HF (16) and were in the amide form at the carboxyl terminus. To obtain peptides with their amino terminus acetylated, we also treated some peptide resin with acetylimidazole prior to cleavage. All peptides were purified in two steps by reverse phase HPLC. In the first step, a solvent system containing a phosphate buffer (pH 3.1) and hexanesulfonic acid (sodium salt) (see above) was used. In the second step, a volatile solvent system containing 0.1% trifluoroacetic acid for further puri- fication and desalting was used. The identity of the peptides was confirmed by sequence analysis on an Applied Biosystem gas phase Sequencer. Peptides were quantified by amino acid compositional analysis (19).

Phosphorylation of Synthetic Peptides-The estimation of kinetic parameters of peptide phosphorylation was done as described else- where (16). Phosphorylation was done with different concentrations of peptide and 0.1 mg of chlorophyll/ml thylakoid membranes, 20 mM NaF, 1 mM MgCl,, 50 mM Tricine (pH 7.8), 200 p M ATP, and 200-300 pCi/pmol [y-"PIATP. Phosphorylation was done in a water bath at 23 "C with illumination by an unfiltered tungsten lamp (200 pE.m-,.s-'). After removal of the membranes by a short centrifuga- tion, the supernatant was passed through an anion exchanger column (AG1-X8, acetate form) in 30% acetic acid (23). Whereas the un- reacted ATP and the free phosphate were adsorbed on the ion exchanger resin, the phosphorylated peptide was not retained. Incor- porated phosphate was measured by Cerenkov counting. Kinetic data were plotted on double reciprocal plots. Apparent K,,, and V,,, values were determined by fitting the data to the Michaelis-Menten equation using the method of least squares.

RESULTS

Isolation and Sequencing of Phosphorylation Sites of LHC ZI-When thylakoid membranes were incubated in the light with [T-"~P]ATP, the 25- and 27-kDa forms of LHC I1 were

17586 Post-translational Modifications of LHC 11

the most conspicuously labeled phosphoproteins (Fig. 1). Lesser amounts of :12P were incorporated into the PS I1 core proteins (8, 32, 34, and 43 kDa) whose phosphorylation sites were sequenced previously (13, 14). Treatment of thylakoids with a variety of proteases released phosphorylated peptides from the membranes, but thermolysin and proteinase K proved most suitable for release of LHC I1 phosphopeptides. The 27-kDa form of LHC I1 was generally more susceptible to proteolytic cleavage than the 25-kDa form (24, 25). By treatment with thermolysin for a limited time, we were able to release phosphopeptides selectively from the 27-kDa form (Fig. 1). On the other hand, extensive treatment with protein- ase K released all radioactivity (data not shown). To isolate phosphopeptides, we removed digested membranes by cen- trifugation and passed the supernatant through a column of Fe:"-iminodiacetyl-Sepharose. Bound phosphopeptides were eluted and fractionated on reverse phase HPLC (Fig. 2). With both proteases, we obtained several major UV-absorbing peaks, all of which were radioactively labeled and therefore contained phosphopeptides.

Five HPLC fractions after thermolytic treatment (Tl-T5) and three fractions after proteinase K treatment (Pl-P3) were examined for amino acid content (Table I). Several fractions (Tl , T2, P1, P2, and P3) were characteristically basic, as expected for the phosphorylation site of LHC I1 (2, 5, 9) and are therefore most likely to be derived from the amino ter- minus of LHC 11. They are unrelated to the amino-terminal phosphopeptides released from purified PS I1 particles (13, 14). We therefore subjected the peptides in fractions T1, T2, and P2 to further sequence analysis. Since we had selectively released the thermolytic phosphopeptides under enzyme-lim- ited conditions (Fig. l), we concluded that the phosphopep- tides contained in the HPLC fractions T1 and T2 originated from the 27-kDa form of LHC 11. On the other hand, protein- ase K is less specific, and, after 48 h of digestion, we obtained a number of different phosphopeptides. All three HPLC peaks (Pl , P2, and P3) showed amino acid compositions which were expected for LHC I1 amino termini, based on sequencing of cloned cab genes of higher plant species other than spinach (2). Whereas P1 is probably formed by further digestion of P2, P3 seems to contain a mixture of different phosphopep- tides. Several of the peaks contained more than one phospho- peptide. This conclusion is based on the amino acid compo- sition and an analysis of the phosphoamino acid content. Whereas T2, P1, and P2 contained only phospho-Thr, we found both phospho-Thr and phospho-Ser in T1 and P3 (data not shown). We also digested with proteinase K subsequent

A 0 LD33 - 92.5 - . 66.2 - 45.0 - 31.0 - And 21.5 - 14.4 - *

0 2 4 9 2 6 0 2 4 9 2 6

Time (h)

FIG. 1. Digestion of spinach thylakoid membranes with thermolysin. "'P-Labeled thylakoids from spinach were treated with thermolysin, and aliquots were separated by sodium dodecyl sulfate- polyacrylamide gel electrophoresis a t different times. A, Coomassie- stained gel; R, autoradiography of the same gel. The position of LHC I1 proteins is marked.

-~ 0 20 40 60 rnin

n

0 20 40 60 rnin

FIG. 2. Separation of phosphopeptides by reverse phase HPLC. A, phosphopeptides released from spinach thylakoid mem- branes by thermolysin and separated with a solvent system containing phosphate and hexanesulfonate (sodium salt). B, phosphopeptides released by proteinase K and separated with a solvent system con- taining heptafluorobutyric acid.

TABLE I Relative amino acid composition of phosphopeptides released from

spinach thylakoids by proteolytic digestion Values were normalized to arginine.

HPLC peak T1 T2 T3 T4 T5 P1 P2 P3

Ala Arg Asx" Glx" GlY Ile Leu LYS Phe Pro Ser' Thr' Val

1.25 1.00 1.05 0.13 1.22 ND ND 3.02 ND 1.10 0.60 0.32 0.17

1.03 1.00 0.09 ND 0.98 ND ND 2.67 ND 0.96 0.39 1.22 ND

0.69

ND 1.00

0.76 0.60 ND ND 0.34 ND 1.18 1.55 0.91 0.63

0.96 1.00 ND 1.60 0.71 ND ND 1.24 ND 1.29 3.27 2.29 0.94

1.47 1.00 2.00 0.81 2.41 0.53 0.80 0.68 0.81 ND 2.60 2.21 1.27

0.08 2.00 0.09 0.11 0.10 ND 0.29 1.17 0.26 ND ND 1.13 1.03

1.02 1.35 2.00 1.00 0.10 0.20 1.02 ND 0.12 1.28 ND ND ND ND 1.05 4.10 ND ND 1.24 1.28 0.93 0.25 0.67 1.05 1.05 0.10

" Asx, Asn or Asp; Glx, Glu or Gln. ND, not detected. Substoichiometric recovery of Thr or Ser is due to incomplete

hydrolysis of phospho-Thr or phospho-Ser. His, Met, and Tyr were not detected. Trp and Cys were not estimated.

to mild thermolytic treatment. Unfortunately, we obtained more fragmentation and generally smaller sized peptides. Peak P2, in particular, was not obtained under these condi- tions.

Phosphopeptide HPLC fractions T1, T2, and P2 were fur- ther chracterized by mass spectrometry. Spectra recorded on the quadrupole Fourier transform instrument (Table 11) in- dicated that fractions T2 and P2 each contained a single peptide. Signals for the corresponding (M + H)' ions were observed at m/z 1108 and 1165, respectively. Fragments re- sulting from the loss of phosphoric acid (98 Da) also appeared in each spectrum. As shown in Fig. 3, the mass spectrum

Post-translational Modifications of LHC 11 17587

TABLE I1 Monoisotopic masses and sequences of the underiuatized (M + H)' and the acetylated Ac-(M + H)'

phosphopeptides in HPLC fractions TI, T2, and P2 Fraction (M + H)' Ac-(M + H)' Seauence

Tla Tlb T2 P2

1107.4 1233.4 Ac-Arg-Lys-Sera-Ala-Gly-Lys-Pro-Lys-Asn 1121.4 1247.4 1108.4

Ac-Arg-Lys-Thr"-Ala-Gly-Lys-Pro-Lys-Asn 1234.4 Ac-Arg-Lys-Thr"-Ala-Gly-Lys-Pro-Lys-Thr

1164.9 1207.2 Ac-Arg-Arg-Thr"-Val-Lys-Ser-Ala-Pro-Gln I' Phosphorylated amino acid residue.

1107 1121

900 1000 1100 1200 m/z

FIG. 3. Mass spectrum of HPLC fraction T1. Spectrum of the underivatized HPLC fraction T1 recorded on the tandem quadrupole Fourier transform mass spectrometer. A 20- to 50-pmol sample in 1 pI of 0.1% triflouroacetic acid was added to a matrix of glycerol/ thioglycerol (1:2).

recorded on fraction T1 contains signals at m/z 1121, 1107, 1023, and 1009. The first two signals correspond to (M + H)' ions for two different peptides, T la and T lb . Loss of phos- phoric acid (98 Da) from these ions accounts for the fragments observed at m/z 1023 and 1009. For each of the above four peptides, the measured molecular weights differ from that calculated from the amino acid compositon data by 122 Da. This is the expected result for peptides containing a single phosphate group (80 Da) on either Thr or Ser and an acetyl group (42 Da) on the amino terminus.

Treatment of the peptides in fractions T1 and T2 with acetic anhydride shifted the observed (M - H)' ions to higher mass by 126 Da. Acetylation under the above conditions occurs only on amino groups in the peptides. Introduction of a single acetyl group increases the molecular mass by 42 Da. The observed mass shift of 126 Da suggested that the three peptides in these fractions (Tla, Tlb, and T2) each contained either a blocked amino terminus and 3 Lys residues or a free amino terminus and 2 Lys residues. Acetylation of peptide in fraction P2 shifted the (M + H)' ion to higher mass by only 42 Da. This result is consistent with the presence of either a single Lys residue and a blocked amino terminus or no Lys residues and a free amino terminus.

Sequence analysis of the above four peptides was performed by the technique of collision-activated dissociation on a triple quadrupole mass spectrometer. This method is particularly useful for sequencing peptides in mixtures (20), for locating sites of phosphorylation (14, 22), and for analyzing samples with blocked amino termini (14). Sample ionization/volatili- zation was performend by either particle bombardment with Cs' ions or by electrospray. The first quadrupole mass filter of the triple quadrupole instrument was employed to selec- tively transmit (M + H)' ions from a particular peptide in the mixture. In quadrupole 2, the selected peptide (M + H)' ions were then subjected to multiple collisions with argon atoms in order to generate fragments resulting from the more or less random cleavage of each amide bond in the sample

molecules. Fragment were then transmitted to quadrupole 3 where they were separated and counted according to mass.

Shown in Fig. 4 is the collision-activated dissociation spec- trum recorded on (H + H)' ions for the acetylated derivative on one of the two peptides, T la , found in fraction T1. The sequence deduced from this spectrum is shown at the top of Fig. 4. Predicted monoisotopic masses for fragment ions of type b and y for this sequence are shown above and below the structure, respectively. Those observed in the spectrum are underlined. Ions of type b all contain the amino-terminal residue plus 1 or more additional residues and have the general structure, H-[NHCH(R)CO],'. Subtraction of m/z values for any two fragments of type b that differ by a single amino acid, NHCH(R)CO, generates a value that identifies the extra amino acid residue present in the larger fragment. Ions of type y all contain the carboxyl-terminal residue plus 1 or more additional residues and have the general structure, H'- [NH,CH(R)CO],OH. Again, subtraction of m/z values for any two fragments of type y that differ by a single amino acid, NHCH(R)CO, generates a value that identifies the extra amino acid residue present in the larger fragment. Additional fragment ions that result from the loss of small molecules such as ammonia (17 Da), water (18 Da), and phosphoric acid (98 Da) often accompany formation of ions of type b and y. Loss of carbon monoxide (28 Da) occurs from fragments of type b but not from fragments of type y.

Signals at m/z 199 and 369 in Fig. 4 are assigned to ions of type b because these ions lose carbon monoxide to form fragments at m/z 171 and 341. The observed m/z values are uniquely characteristic of the sequence, N-AcArg-AcLys. The next ion of type b occurs a t m/z 536 and suffers loss of phosphoric acid (98 Da) to form the strong signal a t m/z 438. The mass difference between the b:, and bz ions is 167 Da, the expected value for a phosphoserine residue. Ions of type b a t m/z 607, 664, and 834 define the next 3 residues as Ala-Gly- AcLys. Note that each of these ions contains the phosphoser- ine residue and each is accompanied by additional fragments resulting from the loss of phosphoric acid (98 Da). The next ion of type b appears at m/z 1101,267 Da higher than the be. The observed mass shift is too high to be accommodated by a single residue, but does match that expected for combination of the two amino acids, Pro and AcLys. Weak or missing signals for ions of type b are often observed in low energy collision-activated dissociation spectra if the corresponding fragments are formed by cleavage of amide bond on either side of Pro residues.

Note that the corresponding ion of type y that contains the last 3 residues appears as a strong signal a t m/z 400. Ions of type y containing Pro at the amino terminus are generally among the most abundant signals in spectra recorded on the triple quadrupole instrument. Subsequent fragmentation of this type y ion generates a series of type b ions, all of which contain Pro at the amino terminus (20). The type bz ion in this series occurs a t m/z 268 (loss of CO at m/z 241) and thus confirms the sequence Pro-AcLys. The mass difference be-

17588 Post-translational Modifications of LHC 11

FIG. 4. Collision-activated disso- ciation mass spectrum. Spectrum re- corded on the triple quadrupole mass spectrometer of the (M + H)+ ion from the acetylated LHC I1 phosphopeptide (m/z = 1107). A 100-200 pmol sample in 1 pl of 6 N HCl in dimethyl sulfoxide was added to a matrix of thioglycerol.

199.1 536.3 U 564.4 M 931.5 W W b,

Ac-Arg Ac-Lys P-Ser Ala Gly Ac-Lys Pro AC-LyS Asn

1035.5 865.4 698.4 627.3 570.3 400.2 303.2 133.1 y,

4

A. .- .. J. L..A 100 150 200 250 300 350

600 900 9 b lobo

I MH* - @ I MH+

.I&. .I 1100 1150 1200 1250 1300

m/z 10io

tween the (M + H)+ ion and the ion of type bs, 132 Da, corresponds to the residue mass of Asn, NHCH(R)CO, plus the carboxyl-terminal " O H group and a proton. Loss of Asn as the residue mass with retention of both the " O H group and the proton on the fragment generates the signal at mlz 1119. We conclude that the (M + H)' ion at mlz 1107 for the underivatized peptide in fraction T1 has the sequence, N- AcArg-Lys-pSer-Ala-Gly-Lys-Pro-Lys-Asn.

Shown in Fig. 5 is the collision-activated dissociation mass spectrum recorded on (M + 2H)'+ ions, mlz 603.5, from the acetylated derivative of the peptide observed in fraction P2. To record this spectrum, sample at the 20-pmol level was loaded onto a CI8 resin contained in a 75-p fused silica column and then eluted with an acetic acidlacetonitrile gradient directly into the mass spectrometer via the electrospray ion source. In the electrospray ionization technique, the number of protons added to the peptide sample is dependent on the number of basic residues contained in the molecule. The acetylated peptide in fraction P2 acquires two protons and affords a strong signal corresponding to the (M + 2H)'+ ion at mlz 603.5.

Shown at the top of Fig. 5 is the amino acid sequence deduced from the fragmentation pattern present in the colli- sion-activated dissociation mass spectrum. Predicted singly charged ions of type b and y for this sequence are shown above and below the sequence. Those observed in the spec- trum are underlined. The first 7 residues in the sequence are deduced from singly charged ions of type b. Fragments that contain the phosphothreonine residue, b3-b7, all lose phos- phoric acid to generate the signals at mlz 438, 537, 707, 794, and 895. Doubly charged ions for b3, b6, and b7 are also observed at mlz 268, 403, and 482. Ions of type y in the spectrum at mlz 147, 244, and 315 facilitate assignment of the last 3 residues in the peptide as Ala-Pro-Gln. We conclude that the peptide in HPLC fraction P2 has the sequence, Ac- Arg-Arg-pThr-Val-Lys-Ser-Ala-Pro-Gln.

Synthetic Peptide Analogs of LHC II Phosphorylation Site- The four peptides shown in Table I1 have a number of

REIATNE AWNDANCE

I x 12 I x 31 x 12 I

I I Y,

60

20

m ux)

600 8M

I x 20 Ix 101 x 100 I

1Mo mlz

FIG. 5. Collision-activated dissociation mass spectrum.

nigan TSQ 70) of the (M + 2H)" ion from the acetylated LHC I1 Spectrum recorded on the triple quadrupole mass spectrometer (Fin-

phosphopeptide (m/z = 1207). A 20-pmol sample was injected into the mass spectrometer by the technique of electrospray ionization.

distinctive features which could, in principle, impact on the rate of phosphorylation (or dephosphorylation). These fea- tures include amino-terminal acetylation, the presence of basic residues on either side of the phosphorylation site, the

Post-translational Modifications of LHC II 17589

replacement of lysine at position 2 (in Tla+b and T2) by arginine (in PZ), and the presence of hydroxyamino acids in position 3, compared with positions 3, 5, and 6 in the pea peptide (R,K)SATTKK (9). To investigate the significance of these features, we synthesized a set of 8 peptide variants of the sequence Ac-RKTAAKPA. That the canonical peptide was indeed phosphorylated by LHC I1 kinase is indicated by the fact that two inhibitors of the light activation of the LHC 11 kinase (diuron and 2,5-dibromo-3-methyl-6-isopropyl-p- benzoquinone (5, 16, 26)) inhibited phosphorylation of the peptide: 95% inhibition by diuron and 90% inhibition by 2,5- dibromo-3-methyl-6-isopropyl-p-benzoquinone (result not shown). The canonical peptide resembles the peptides in Table 11, but was simplified to permit synthesis of a set of derivatives with specific structural alterations without the need for a large number of control peptides. In addition, only one hydroxyamino acid was included in each peptide to avoid the need for sequencing to identify the phosphorylated resi- due(s).

The eight peptides, in both acetylated and nonacetylated forms, were incubated with thylakoids under phosphorylation conditions and apparent &(peptide) and V,,, values were calculated (Table 111). The use of peptides to study phos- phorylation in thylakoid membranes has already been de- scribed (16). Peptides l and 2 were similar to the amino terminus of a class I LHC I1 (Arg-Lys motif) and a class I1 LHC I1 (Arg-Arg motif), respectively. The kinetic parameters for peptides 1 and 2 in acetylated and nonacetylated forms permit two conclusions. First, replacement of lysine by an arginine at position 2, which is a characteristic difference between class I and class I1 LHC IIs, had little effect on phosphorylation: K, was unchanged, and V,,, was only slightly reduced. Second, acetylation had little influence on phosphorylation.

With the six remaining peptides, the effects of two kinds of sequence changes were examined (i) displacement of the threonyl residue, and (ii) replacement of basic residues on either side of the threonine. When threonine was moved from position 3 to position 4 or 5, K,(peptide) increased and V,,, declined (compare peptides 1, 3, and 4); both factors tend to reduce the initial velocity of the reaction. This effect was more marked when threonine was moved to position 5 com- pared with position 4 and was intensified by acetylation. Replacement of the lysine on the carboxyl-terminal side of threonine was found to increase &(peptide) markedly (com-

pare peptides 1 and 6). Replacement of the amino-terminal arginine by alanine (peptide 7) produced an even more marked increase in &(peptide). Replacement of the Arg-Lys motif by Ala-Ala (peptide 8) reduced phosphorylation to below a measurable rate. Since &(peptide) is a measure of the affinity between kinase and substrate, we conclude that the 3 basic residues flanking the LHC I1 phosphorylation site play a major role in binding to the protein kinase.

DISCUSSION

From spinach thylakoids we have isolated four LHC I1 phosphopeptides which have been sequenced by tandem mass spectrometry. All four peptides were blocked by N-acetylation of amino-terminal arginines. The acetyl group was present prior to chemical acetylation of lysine residues and indicates that each phosphopeptide was derived from the amino ter- minus of the corresponding thylakoid-bound protein. Fur- thermore, the fact that the four phosphopeptides differed in amino acid sequence indicates that spinach contains at least 4 distinct LHC I1 cab genes. One of these genes appears to have been already cloned and sequenced: the T2 peptide corresponds to the amino terminus of the cab gene reported by Mason (27) and by Wedel (cited in Ref. 8).

The amino terminus of only one other LHC I1 has been sequenced at the protein level (9). In that case (a pea protein), the order of the first 2 residues (arginine and lysine) could not be determined, because the tryptic dipeptide (R,K) was blocked against Edman degradation (9). Our data suggest that the pea LHC I1 had the amino terminal sequence Ac- RKSATTKK, corresponding to a pea gene cloned by Coruzzi et al. (10).

It is usually assumed that mature LHC I1 is generated by cleavage of the precursor immediately in front of a conserved methionine in the sequence -Met-Arg-(Arg/Lys)-. It is not clear how this view (see, for example, Refs. 2, 8, and 27) has gained currency, because the data of Mullet (9) suggest that the amino-terminal residue in pea is not methionine but either arginine or lysine. Our data (and our interpretation of Mullet’s data) strongly indicate that cleavage occurs between the con- served methionine and the conserved arginine.

N-Acetylation of proteins has emerged as an intriguing feature of spinach photosystem 11. In addition to being found in four LHC I1 molecules, it is also found in three out of four phosphoproteins of PS I1 cores (13, 14). The three proteins are the products of the chloroplast psbA, -C, and -D genes,

TABLE I11 Apparent kinetic constants (measured at 0.2 m M ATP) for phosphorylation of synthetic peptide substrates by

thylakoid-bound LHC II kinase The peptides were phosphorylated as described under “Experimental Procedures.” Values are from at least two

independent experiments. Amino terminus

Peptide Free Acetylated

K* Vmaxb Kme Vmnx 1) Arg-Lys-Thr-Ala-Ala-Lys-Pro-Ala 2) Arg-Arg-Thr-Ala-Ala-Lys-Pro-Ala 3) Arg-Lys-Ala-Thr-Ala-Lys-Pro-Ala 4) Arg-Lys-Ala-Ala-Thr-Lys-Pro-Ala 5) Arg-Lys-Thr-Ala-Ala-Ala-Pro-Ala 6) Arg-Lys-Ala-Ala-Thr-Ala-Pro-Ala 7) Ala-Lys-Thr-Ala-Ala-Lys-Pro-Ala 8) Ala-Ala-Thr-Ala-Ala-Lys-Pro-Ala

K, in mM peptide. V,,, in nmol/mg of chlorophyll/h. NE, not estimated. ND, not detected.

0.31 f 0.05 0.28 f 0.08 0.41 f 0.03 0.47 f 0.03 0.88 f 0.18 0.74 f 0.14 1.27 f 0.16

ND

47 f 1.9 37 f 2.5 21 f 1.5 21 f 1.7 46 f 6.0 21 f 6.0 48 f 8.1

ND

0.32 C 0.07 0.28 0.08 0.81 f 0.02 1.24 f 0.07 0.91 f 0.14

NE‘ 1.51 f 0.23

ND

43 f 9.4 37 f 9.3 14 f 1.6 16 f 2.0 26 f 1.5

NE 32 f 0.7

ND

17590 Post-translational Modifications of LHC II and each begins with N-acetyl-0-phosphothreonine. The 32- kDa psbA product (protein Dl) and the 34-kDa psbD product (protein D2) comprise the reaction center of PS 11, the 43- kDa psbC product is a proximal antenna protein, and the 25- and 27-kDa LHC 11s constitute the distal antenna proteins. I t is unlikely that N-acetylation is required for phosphoryla- tion because (i) the fourth phosphoprotein of PS I1 cores (the psbH gene product) contains the amino-terminal sequence alanyl-phosphothreonyl which is not acetylated, and (ii) as we show here, acetylation of synthetic peptide analogs of the phosphorylation site of LHC I1 does not enhance phosphoryl- ation. It is not known whether the same acetyltransferase acts on LHC I1 to acetylate arginine and on PS I1 core proteins to acetylate threonine. In contrast, no PS I protein is known to be blocked by amino-terminal acetylation.

The four spinach LHC I1 molecules examined here are phosphorylated on the 3rd residue, which may be serine or threonine. The finding that only one out of four peptides contains phosphoserine is consistent with the report that phosphoserine represents a minority of the 'lP in spinach LHC I1 (28). A survey of most known higher plant LHC I1 gene sequences (2) reveals that all but one of these sequences predict either serine or threonine at the third position. The exception is a LHC I1 of tomato (29), which contains alanine in the third position. Thus, the vast majority of cab genes would appear to code for phosphorylatable LHC 11s.

However, it is not necessarily true that phosphorylated LHC I1 will always prove to carry the phosphoryl group on the 3rd residue. Mullet (9) showed that the pea phosphoryla- tion site was (R,K)SATTKK, with phosphate carried on position 5 or 6 or both (he could not distinguish among these possibilities) but with no phosphate on the serine in position 3. Our present data on the phosphorylation of synthetic peptide analogs of LHC I1 establish that LHC I1 kinase is able to phosphorylate peptides on threonine in position 3, 4, or 5 , but the data do not explain Mullet's finding that threo- nine in position 5 or 6 is phosphorylated in preference to serine in position 3. Nevertheless, Mullet's report is undoubt- edly correct because when we supplied thylakoids with a synthetic peptide containing the sequence RKSATTKK, j2P was incorporated into threonine a t position 5 but was not found in the serine at position 3 (16). However, a peptide containing the sequence RKSASSKK was phosphorylated preferentially a t position 3 (16). Thus, the site of phosphoryl- ation of LHC I1 and synthetic peptide analogs is a complex function of the number and type of hydroxyamino acids present.

Phosphorylation depends also on the presence of basic residues. The spinach phosphopeptides sequenced here and the pea phosphopeptide isolated by Mullet (9) contained basic residues on both sides of the phosphorylated residue. The data presented here on a set of synthetic peptide analogs of the LHC I1 phosphorylation site establish that both groups of basic residues are important. Basic residues also play an important role in substrates for certain other protein kinases. The CAMP-dependent protein kinase is one of several well studied serine/threonine protein kinases of animal cells. In- vestigations with synthetic peptides indicate that the CAMP- dependent kinase requires basic residues on the amino-ter- minal side of the phosphorylation site (30,31). Protein kinase C requires basic residues on both sides of the phosphorylation site (32). However, casein kinase requires flanking acidic residues (33). On this very limited basis, LHC I1 kinase would appear to be most similar to protein kinase C.

Pre-LHC I1 molecules are encoded by two classes of cab genes (2). The consensus sequence of type I LHC I1 is Arg-

Lys-(Thr/Ser)-(2 residues)-Lys-, whereas the consensus se- quence of type I1 LHC I1 is Arg-Arg-(Thr/Ser)-(1 residue)- Lys-. Our data are consistent with this division. The three peptides generated by thermolysin treatment are type I se- quences and are derived from the 27-kDa LHC 11. The peptide released by proteinase K appears to be derived from the 25- kDa protein and is of type 11. It has been reported that the 25-kDa LHC I1 is much more rapidly phosphorylated than the 27-kDa LHC I1 (34, 35). That difference in rate does not seem to be attributable simply to the replacement of lysine by arginine at position 2, because, as we show above, synthetic peptides differing in this respect show no significant kinetic difference during phosphorylation by thylakoids. Perhaps the closer spacing of basic residues in type I1 LHC 11s is respon- sible for their more rapid phosphorylation. I t would be in- structive to use synthetic peptides to characterize and perhaps purify the phosphatase responsible for LHC I1 dephosphoryl- ation ( 5 ) .

In summary, our data point to the existence of a three-step postsynthetic modification scheme for LHC 11: (i) proteolytic processing at a conserved Met-Arg peptide bond, (ii) acety- lation of the revealed N"-amino group of Arg, and (iii) phos- phorylation of a neighboring Thr/Ser residue that is usually 2 residues downstream from the processing site. Basic residues flanking the phosphorylation site are important for recogni- tion by the LHC I1 protein kinase. We have conducted site- directed mutagenesis of several residues in and around the cleavage site and examined the effect of these mutations on the ability of intact chloroplasts to process pre-LHC II.2 While replacement of the conserved Met by Val forces the processing protease to cleave at an unidentified site slightly downstream, replacement of the conserved Arg by either His or Leu has no discernable effect on processing. Thus, the arginine residue is conserved more for its role in kinase recognition than for its role in protease recognition.

REFERENCES

1. Chitnis, P. R., and Thornber, J. P. (1988) Photosynth. Res. 16,

2. Demmin, D. S., Stockinger, E. J., Chang, Y. C., and Walling, L.

3. Keegstra, K., and Olsen, L. J. (1989) Annu. Reu. Plant Physiol.

4. Yalovsky, S., Schuster, G., and Nechushtai, R. (1990) Plant Mol.

5. Bennett, J. (1984) Physiol. Plant. 60, 583-590 6. Lamppa, G. K., and Abad, M. S. (1987) J. Cell Biol. 105, 2641-

7. Robinson, C., and Ellis R. J. (1984) Eur. J . Biochem. 142, 337-

8. von Heijne, G., Steppuhn, J., and Herrmann, R. G. (1989) Eur.

9. Mullet, J. E. (1983) J. Bid. Chem. 258, 9941-9948

41-63

L. (1989) J. Mol. E d . 29, 266-279

Plant Mol. Biol. 40, 471-501

Biol. 14, 753-764

2648

342

J. Biochem. 180, 535-545

10. Corruzzi, G., Broglie, R., Cashmore, A,, and Chua, N.-H. (1983)

11. Hoober, J. K., Millington, R. H., and D'Angelo, L. P. (1980) Arch.

12. Burgi, R., Suter, F., and Zuber, H. (1987) Biochim. Biophys. Acta

13. Michel, H. P., and Bennett, J. (1987) FEBS Lett. 212, 103-108 14. Michel, H., Hunt, D. F., Shabanowitz, J., and Bennett, J. (1988)

15. Bennett, J., Shaw, E. K., and Bakr, S. (1987) FEBS Lett. 210,

16. Michel, H. P., and Bennett, J. (1989) FEBS Lett. 254, 165-170 17. White, I. R., O'Donnell, P. J., Keen, J. N., Findlay, J. B. C., and

18. Bhalla, P., and Bennett, J. (1987) Arch. Biochem Biophys. 252,

J. Biol. Chem. 258, 1399-1402

Biochem. Biophys. 202, 221-234

890, 346-351

J. Bid. Chem. 263, 1123-1130

22-26

Millner, P. A. (1990) FEBS Lett. 269, 49-52

2149-2154

' W. Buvinger and J. Bennett, unpublished results.

Post-translational Modifications of LHC 11 17591

19. Hirs, C. H. W. (1983) Methods Enzymol. 91, 3-8 20. Hunt, D. F., Yates, J. R., 111, Shabanowitz, J., Winston, S., and

6237 Hauer, C. R. (1986) Proc. Natl. Acad. Sci. U. S. A. 83, 6233-

21. Hunt, D. F., Shabanowitz, J., Yates, J. R., 111, Zhu, N. Z., Russell, D. R., and Castro, M. E. (1987) Proc. Natl. Acad. Sci. U. S. A.

22. Erickson, A. K., Payne, D. M., Martino, P. A., Rossomando, A. J., Shabanowitz, J., Weber, M. J., Hunt, D. F., and Sturgill, T. W. (1990) J. Bid. Chem. 265, 19728-19735

23. Kemp, B. E., Benjamini, E., and Krebs, E. G. (1976) Proc. Natl.

24. Michel, H. P., Schneider, E., Tellenbach, M., and Boschetti, A.

25. Suss. K. H.. Schmidt, 0.. and Machold, 0. (1976) Biochim.

84,620-623

A c u ~ . S C ~ . U. S. A. 73, 1083-1042

(1981) Photosynth. Res. 2, 203-212

~iOphys. A& 448, io3-i13 26. Bennett. J.. Shaw. E. K., and Michel, H. P. (1988) Eur. J.

Biochem. '171, 95-100

27. Mason, J. G. (1989) Nucleic Acids Res. 17, 5387 28. Lucero, H. A,, Cortez, N., and Vallejos, R. H. (1987) Biochim.

Biophys. Acta 890, 77-81 29. Pichersky, E., Bernatzky, R., Tanksley, S. D., Breidenbach, R.

B., Kausch, A. P., and Cashmore, A. R. (1985) Gene (Amst.) 40,247-258

30. Kemp, B. E., Graves, D. J., Benjamini, E., and Krebs, E. G. (1977) J. Biol. Chem. 252,4888-4894

31. Feramisco, J. R., Glass, D. B., and Krebs, E. G. (1980) J. Biol. Chem. 255,4240-4245

32. House, C., Wettenhall, R. E. H., and Kemp, B. E. (1987) J. Biol. Chem. 262,172-777

33. Meggio, F., Marchiori, F., Borin, G., Chessa, G., and Pinna, L. A. (1984) J. Biol. Chem. 259, 14576-14579

34. Islam, K., and Jennings, R. C. (1985) Biochim. Biophys. Acta 810,158-163

35. Islam, K. (1987) Biochim. Biophys. Acta 893,333-341